Data Repository

We have used the dataset of Joint EVS/WVS 2017-2020, allowing for comparisons of tax morality in more than 79 countries worldwide, applying machine learning variable selection methods suited to categorical variables (Chi-Squared and mutual information).
The empirical results revealed that variables like: religious denomination, faith in God and the importance of God, together with the level of trust in people from other religion, confidence in churches have been connected with a significantly increased the extent of tax morale, being the most relevant variables of religiosity.

本研究采用2017-2020年联合欧洲价值观调查（European Values Study, EVS）与世界价值观调查（World Values Survey, WVS）数据集，该数据集支持对全球79个以上国家的税收道德开展比较分析，并运用适配分类变量的机器学习变量选择方法（卡方检验与互信息）。
实证结果表明，宗教教派、对上帝的信仰及上帝的重要性，连同对其他宗教信徒的信任程度、对教会的信心等变量，均显著提升了税收道德感，是宗教性维度中最具关联性的变量。